J. Am. Chem. SOC.1994,116, 7557-7563
7557
Kinetic Isotope Effect Study of Transition States for the Hydrolyses of a- and P-Glucopyranosyl Fluorides Yulei Zhang, Jeyashri Bommuswamy, and Michael L. Sinnott' Contributionfrom the Department of Chemistry (M/Cll I ) , University of Illinois at Chicago, 845 West Taylor Street, Chicago, Illinois 60607-7061 Received January 14, 1994'
Abstract: The hydrolyses of a- and 8-glucopyranosyl fluorides at near-neutral pH have been studied. The water (buffer-independent) reactions were characterized by values of AH*of 96 f 4 and 88 5 kJ mol-' and of AS* of -37 f 12 and -38 f 17 J mol-' K-l, respectively, and by multiple kinetic isotope effects measured by the isotopic quasiracemate method. These effects, expressed as klight/kbcavy for the a-fluoride at 80 OC and the &fluoride at 50 OC, were, respectively, for a D (Cl), 1.142 and 1.086, for OD (C2), 1.065and 1.030, for ring 180,0.984 and 0.985, and for anomeric 13C,1.032 and 1.01,; a y D (C5) effect of 0.980 was also measured for the a-compound. A transition-state structure for the hydrolysis of the a fluoride, involving the ring in a flattened chair conformation, and an "exploded" sN2 dispositionof water and fluoride ion about theanomeric center were located from BEBOVIB-IV and the initial structure, parameter settings, and procedure used by Tanaka et al. (Tanaka, Y . ;Tao, A.; Blanchard, J. S.; Hehre, E. J. J. Biol. Chem. under review) in their study of the glucoamylase-catalyzed hydrolysis of this compound. The effects for the 8-fluoride are consistent with a more SNl-like transition state in which the sugar ring is again in a flattened 4 C ~ conformation. In the presence of 2.0 M sodium azide, when the reaction of the a-fluoride becomes a classic sN2 reaction (Banait, N. S.;Jencks, W. P. J . Am. Chem. SOC.1991,113, 7951), the 13Ceffect increases to l.O& but the a D effect also increases modestly (to l.169). The reaction of the &fluoride in 0.3 M succinate buffer is largely (formally) a bimolecular reaction of the acidic form of the buffer; effects are as follows: aD, 1.1 lo; BD, 1.059; yD, 0.981; ring l*O, 0.988; anomeric l3C, 1.064. The apparent promotion of ANDN reactions of the ,&fluorideby succinate monoanion rather than dianion suggests that the microscopic reaction is the catalysis by neutral succinic acid of the formation of 1,2anhydroglucose from the monoanion of the glucosyl fluoride.
*
Introduction The mechanisms of the hydrolysis and solvolysis of glycosyl derivatives are of interest as the simplest nonenzymic counterpart of the glycosyl transfers important in biology,' where the formation and cleavage of glycosidic linkages is important metabolically, structurally, an! in terms of information transfer? The involvement of species with many of the properties of glycopyranosyl cations in the hydrolysis and solvolysis of glycopyranosyl derivatives has long been recognized,3 but a full understanding of the conformation of these species and the timing of the making and breaking of bonds to the reaction center has yet to be achieved. For many years discrete glucopyranosylcationswere considered to be intermediatesin hydrolysis and solvolysis of glucuopyranosyl derivatives, but measurements of the lifetimes of other oxocarbonium ions4 cast doubt on the real existence of the less stable glucosyl cation as a solvent-equilibrated intermediate even in water. The discovery that the apparently sN1 reactions of methoxymethyl derivativeswere in fact preassociation reactions,5 coupled with the lesser reactivity of glucosyl derivatives, reinforced the probabilitythat the reactions of glucosyl derivatives necessarily involved the participation of the incoming nucleophile. This is certainly the case in mixtures of ethanol and trifluoroethanol,6 but it now appears that the lifetimes of oxocarbonium ions are relatively insensitive to polar effects of the type responsible for
* To whom correspondenceshould be addressed. Abstract published in Advance ACS Abstracts, July 15, 1994. (1) Review: Sinnott, M. L. Chem. Reu. 1990, 90, 1171. (2) Review: Radcmacher, T. W.; Parekh, R. B.; Dwek, R. A. Annu. Reu. Biochem. 1988,57, 785. (3) Reviews of older literature: (a) Capon, B. Chem. Reu. 1969,69,407. (b) Cordes, E. H.; Bull, H. G. Chem. Reu. 1974, 74, 581. (4) (a) Young, P. R.; Jench, W. P. J. Am. Chem. SOC.1977, 99, 8238. (b) McClelland. R. A.; Ahmad, R. J. Am. Chem. SOC.1978,100, 7027. (5) (a) Craze, G. A.; Kirby, A. J.; Osborne, R. J. Chem. Soc., Perkin Tram. 2 1978,357. (b) Knier, B. L.; Jencks, W. P. J . Am. Chem. Soc. 1980, 102,6789. (6)Sinnott, M. L.; Jench, W. P. J . Am. Chem. SOC.1980, 102, 2026. 6
the lower reactivity of glucosyl derivatives compared to tetrahydropyranyl derivatives? so that the estimated lifetime of the glucopyranosyl cation in water is longer than that of the methoxymethyl cation. In accord with this, although there are salt effects on the hydrolysis of glycosyl derivatives with leaving groups which are equatorial in the preferred ground-state conformation, these effects parallel the basicity, rather than the nucleophilicity, of the anion and the neutral nucleophile thiourea has no effect.8 Recently, however,Banait and Jencksg have shown that a-glucopyranosyl fluoride, in which the leaving group is axial, can undergo bimolecular nucleophilic displacements in water, but only with anionic nucleophiles. General base catalysis of the attack of a neutral nucleophile by an anionic base can also be observed. Taken together, these data suggest that the glucosyl cation has a real existence but is not in contact with an anionic nucleophile or leaving group. The question of the conformation of the pyranose ring at the transition state for hydrolysisof glucosyl derivatives can therefore be addressed, in some cases, without having toconsider the position of the incoming nucleophile. A multiple kinetic isotope effect study of the acid-catalyzed hydrolyses of methyl a-and 8-glucopyranosides revealed that the likely conformation of the pyranose ring in the transition state for the hydrolysis of the &compound was a flattened 4C1 chair conformation, whereas that for the hydrolysis of the a-compound was a flattened IS3 skew conformation.s These results stood in flat contradiction to the predictionsof the antiperiplanar lone pair hypothesis (ALPH), that the a-anomer should react through the ground-state 4C1 conformation but the @-anomerwould be forced to adopt a boat conformation, in order to fulfill a supposed "stereoelectronic" requirement that one of the oxygen sp3 lone pairs of electrons (7) Amyes, T. L.; Jencks, W. P. J. Am. C k m . Soc. 1989, I l l , 7888. (8) Bennet, A. J.; Sinnott, M. L. J . Am. Chem. Soc. 1986, 108, 7287. (9) (a) Banait, N. S.;Jench, W. P. J . Am. Chem. Soc. 1991,113,7951. (b) Banait, N. S.; Jench, W. P. J. Am. Chem. Soc. 1991,113, 7958.
OOO2-7863/94/1516-7557$04.50/00 1994 American Chemical Society
7558 J. Am. Chem. SOC.,Vola116. No. 17, I994
should be antiperiplanar to the leaving group.lOJ1 They were however readily understood on the basis of the reverse anomeric effect of the protonated methoxy leaving group, which reinforced the normal preference of the equatorial anomer to adopt the 4C1 conformation but forced the initially axial anomer to adopt the same skew conformation as a-D-glucopyranosylpyridinium ion in solution.12 On this basis it would be predicted that the hydrolyses of both anomers of glucosyl derivatives with an anionic leaving group, which would experiencea direct, rather than a reverse, anomeric effect, would go through transition states in which the pyranose ring approximates to the 4C1conformation of the substrate in its ground state. Such processes are known with fluoride,l4 2,4dinitrophenolate,lsand dihydrogenphosphate16 as leaving groups. We therefore now report a multiple kinetic isotope effect study of the hydrolyses of a-and 8-glucopyranosylfluorides, analogous to that previously reported for the Oglycosides, except that a leaving group kinetic isotope effect requires special techniques:" sites of isotopic substitution are shown in structure I. The kinetic
isotope effects are measured by the isotopic quasi-racemate method, in which the optical rotation of a solution containing equal concentrations of the D-enantiomer highly enriched in the isotope of interest, and of the L-enantiomer, is followed during the course of the reaction.18 (This method therefore cannot be used with glucosyl2,4-dinitrophenolates,which yield the intensely colored 2,4-dinitrophenolateanion; glucosyl phosphates, though they could yield a leaving group kinetic isotope effect, are unsuitable because of competitive phosphorus-oxygen fission.)
Experimental Section Materials. D-Glucose, L-glucose, [2-ZH]-~glucose, [ l-13C]-~glucose (99 atom W I3C), and 97 atom W H2180 were purchased from Aldrich Chemical Co., Milwaukee, WI. 1,2,3,4,6-Penta-O-acetyl-[l - I Z H ] - ~ glucose was made by the method of Berven and Withers.I9 Deuterium and ' 8 0 at C5 were introduce as described previously:8 the key intermediate I1 was reduced with LiAID4 to introduce deuterium. To introduce I*O, the ketone was exchanged with HzI80 (97 atom 96 180)four times, and the exchange was followed by infrared spectrometry ( v c - 1 6 0 1736 cm-I, v c ' 1 1 0 1711 cm-I). After the final exchange, the I80-labeled ketone, which we estimate to be >95% C=I8O from the complete absence of any ( 10) (a) Deslongschamps,P. Stereoelectronic Effecrs in OrganicChemistry;
Pergamon: Oxford, 1983; p 39 especially. (b) Kirby, A. J. The Anomeric Effect and Related Stereoelectronic Effects a? Oxygen; Springer-Verlag: Heidelberg, Berlin, New York, 1983. (11) Thecasehasbffinmadethatthebodyofdatathatledtotheformulation of ALPH is merely the manifestation of least motion effects. Least motion effects are small and elusive and are not observed in reactions with very late or very early transition states. For a detailed discussion of the problem, see: Sinnott, M. L. Ado. Phys. Org. Chem. 1988, 24, 113-207. (12) Hosie, L.; Marshall, P. J.;Sinnott, M. L.J. Chem.Soc., Perkin Trans. 2 1984, 1121. (13) Andrews, C. W.; Fraser-Reid, B.; Bowen, J. P. J. Am. Chem. SOC. 1991,113, 8293. (14) Jones, C. C.; Sinnott, M. L.J . Chem. Soc., Chem. Commun. 1977, 767. (15) Cocker, D.; Sinnott, M. L. J . Chem.Soc., Perkin Trans.2 1975,1391. (16) OConnor, J. V.; Barker, R. Carbohydr. Res. 1979, 73, 227. (17) The longest-lived isotope of fluorine, apart from '9F, is I8F (tll2 1.87 h), but kinetic isotope effects arising from this isotope have been measured (Matsson, 0.;Persson, J.; Axelsson, B. S.;Lingstram, B. J . Am. Chem. Soc. 1993, 115, 5288). A study of 18F effects in the enzymic and nonenzymic hydrolyses of glycosyl fluorides is in hand. (18) (a) Bergson, G.; Matsson, 0.;SjBberg, S.Chem. Scr. 1977, 11, 25. (b) Nadvi, N. S.;Robinson, M. J. T. Abstracts, 4th InternutionalSymposium on Physical Organic Chemistry; The Chemical Society: London, 1978; C1, p 141. (c) Tencer, M.; Stein, A. R.Can. J . Chem. 1978, 56, 2994. (19) Berven, L.A,; Withers, S.G. Carbohydr. Res. 1986, 156,282.
Zhang et al. 0
v O = l 6 0 and FAB mass spectrometry of the derived fluorides,was reduced immediately. Glucose and pentaacetylglucose were converted to the crystalline tetraacetyl a- and @-glucopyranosylfluorides (mp 122-1 24 and 79-8 1 OC, respectively) by literature procedures." Deuterium incorporation was confirmed by ['HI NMR (at 400 MHz) in CDCl3, the anomeric proton at 6 5.74 (q, JF-HI53 Hz, JHI-H~ 3 Hz) in the a-fluoride and 5.40 (9, JF-HI52 Hz, JHI-HZ6 Hz) in the @-fluoridedisappearing in the 1-deuteriatedcompounds and becoming a doublet in the 2-deuteriated compounds, the 2-proton at 6 4.94 (sextet, JHZ-F 24 Hz, JHZH~ 10 Hz) in the a-fluoride and 6 5.12 (sextet, JF-H~ 12 Hz, JHZ-H~ 9 Hz) in the @-fluoridedisappearing in the 2-deuteriated compounds and becoming a quartet in the 1-deuteriated compounds, and the multiplet at 6 4.18 in the a-fluoride and 6 3.92 in the @-fluoridedisappearing in the deuterated compounds. The tetra-O-acetylglucosyl fluorides were deacetylated with potassium cyanide in methanol.21 All labeled a-glucosyl fluorides were crystalline materials (mp 118-120 "C). Due to the difficulties encountered in crystallizing the labile @-fluorides,rather than attempting to crystallize them, immediately after deacetylation, they were dried under vacuo, and after testing by NMR, they were used immediately for kinetic purposes. With small isotope effects for the 8-compound, it was virtually impossible to get the system balanced and so equal quantities of the crystalline tetraacetates were weighed out and deacetylated as a mixture. Kinetic Measurements. First-order rate constants for hydrolyses of aand @-glucosylfluoride, or for mutarotation of glucose, were obtained from the change in rotation at 405 nm of buffered solutions in a 1-dm, 1-mL jacketed cell in a JASCO DIP 370 polarimeter equipped with a printer. Water thermostated to f O . l OC was passed through the cell jacket from a Fisher Scientific temperature controller. Some measurements were made by measurement of fluoride ion release: it was confirmed in representativecases that rateconstants measured by the two monitoring systems were identical. Rate constants were obtained by nonlinear least squares fitting of time courses to eq 1 using the nonlinear least squares fitting program Kuleidugraph (version 20) (Abelbeck software, bought from Synergy (PCS Inc.)) run on a Macintosh SE30 personal computer.
Isotope Effects. Experimental time courses of optical rotation of an isotopic quasi-racemate in the course of a reaction were fitted to eq 2,
The optical rotation change for complete reaction for the light isotopomer (A) was measured in a separate experiment. For the a-fluoride in 1.0 mL (3.0 mg) in a 1-dm pathlength cell, it was 0.310° on hydrolysis and 0.600° for the reaction with 2.0 M sodium azide, and for @-fluoride(3.0 mg) on a hydrolysis, it was O.lOOo. Despite the buffer acceleration of these reactions, A remained unaltered, indicating that any nucleophilic substitution product decomposed on a time scale fast compared to hydrolysis. For small effects, particularly with the @-fluoride,up to 6 mg of each enantiomer was used. In these cases it was confirmed that the pH after reaction remained in the pH-independent region (pH 4.0-7.OZ2). The final pH values were in the region 5.7-6.0. The isotopic effect was then calculated as &L/&H. Fits of experimental data to eq 2 are very ill-conditioned, and data were accepted only when a maximum or a minimum was observed and when the size of the effect calculated by more sophisticated procedures corresponded approximately to that calculated from II = A6k/2.7&, where ll is the amplitude of the maximum or minimum and 6& is the difference in rate constants between light and heavy isotopomers. The procedure depends critically on A and B not being completely indeterminate: the initial estimates of A and B input into Kaleidagruph were those calculated from the amount of each (20) Micheel, F.; Klemer, A. Adv. Carbohydr. Chem. 1961, 16, 85. (21) Herzig, J.; Nudelman, A.; Gottlieb, H. E.;Fischer, B. J . Org. Chem. 1986, 51, 727. (22) Konstantinidis, A.; Sinnott, M. L. Biochem. J. 1991, 279, 587.
-
-
1.00
VI
-
8.00
.
6.00
-
4.00
.
2.00 0.00 100
-
J. Am. Chem. SOC.,Vol. 116, No. 17, 1994 7559 1.2 1 0 . ~ L
1.0 10.3 Y
6.0
. .
4.0 104
.
8.0
. 11 e
2.0
0.0 100
0
0.1
0.2
0.3
Cone.(
M
I
-0.1
0.5
0.4
0
I
0.2
0.1
0.3
0.5
0.4
Cone.( M
)
)
Figure 1. Effect of succinate buffer concentration on the hydrolysis of a-glucosyl fluoride at 80 OC, pH 6.0, Z = 1.0 M (NaC104).
Figure 2. Effect of succinateconcentration on the hydrolysisof j3-glucosyl fluoride at 50 OC, pH 6.0, I = 1.0 M (NaC104). ( 0 )In Bis-Tris buffer at 50 OC, pH 6.0, Z = 1.0 M (NaC104).
antipode in the reaction mixture, and refined values were within 10% of initial estimates. The precision within each individual run was characterized by the standard deviation of optical rotation ua and the effect of fluctuation of optical rotation a on the kinetic isotope effect, obtained from fitting data with eq 2. For a series of measurementsof ai (i = 1,2, ......,N), standard deviation of optical rotation a is
Table 1. First-Order Rate Constants for Hydrolyses of
(3)
Where (6a)i = ai - a p . t
apl
Glucopyranosyl Fluorides in 50 mM Sodium Acetate Buffer, I = 1.0 M (NaC104) T ("c) k, (S-I) k6 (s-') AH $6 = 88 & 5 kJ mol-' 31.0 5.83 X AS $8 = -38 f 17 J mol-' K-I 37.0 40.0 44.0 50.0 55.0 60.0 70.0 80.0
1.17 X lo-' 8.42 X 1od 2.19 X l C 5
2.75 X 10-4 3.98 X 10-4 8.79 X lo-'
8.91 X 2.31 X lo-' 5.46 X lo-'
* a = 96 & 4 kJ mol-' AS ,* = -37 f 12 J mol-' K-I
AH
is the point on best fitting curve at time
= i8t.
The effect of fluctuation of optical rotation a on the kinetic isotope effect, obtained from fitting data to eq 2, was estimated from eq 4:
Table 2. Comparison of the Rates of Hydrolysis of the Glucosyl Fluorides and Mutarotation of Glucose
buffep fluoride T('C) a 80 0.2 M succinate 0.2 M succinate a 60 +2MNaN3 a 50 j3 50 0.3 M succinate 0.3 M Bis-Tris j3 50
kmut
(s-l)
0.088 0.033 0.014 0.015 0.0031
khydro (S-')
0.000 55 0.OOO23 0.000083 0.000 69 O.OOO31
khydrolkmut
0.006 0.007 0.006 0.047 0.1
At pH 6.0, I = 1.0 M (NaClO4). where u, is the standard deviation of the kinetic isotope effect. S is the ) , lAt It INAt. The assumption of kinetic isotope effect ( k ~ l k ~and 8 k 5~ k~ was used in the derivationof eq 3. The second and third terms contribute most of for a nonbalanced run.
Results and Discussion
Nature of the Hydrolysis Reactions. Much of this work was initiated before we were aware of the the work of Banait and Jencks, which indicated that nucleophilicgaand general acid- and base-catalyzed9b processes contributed to the rate of liberation of fluoride ion from a-glucopyranosyl fluoride, but only with anionic reagents. On the basis of previous findings that the nature of the salt did not effect hydrolyses of various glycopyranosyl derivatives,8 we used high concentrations of succinate buffer to neutralize the HF liberated in the hydrolysis. In the event, as is seen from Figure 1, this unfortunate choice did not affect the validityof the results with the a-fluoride: even a t 0.3 M succinate buffer, only -10% of the reaction goes through the buffercatalyzed process. With the &fluoride, however, the dependence on buffer concentration was much more marked (Figure 2) and it is clear that most of the reaction at 0.3 M succinate buffer proceeded
through buffer-catalyzed processes. We therefore performed an additional set of measurements on the @-fluorideusing neutral Bis-Tris as a buffer. In Table 1 are gathered together first-order rate constants for the water reaction of both fluorides determined in low concentrations of acetate buffer. The much higher energy of activation for the a-fluoride means that the a/@ratio is very temperature dependent. Theslightly negative entropies of activation classically suggest bimolecular processes, although the validty of this criterion in reactions in water in which strongly solvated ions aregenerated is questionable: the entropy for the @-fluorideis similar to that for the galacto compound reported p r e v i ~ u s l y . ~ ~ In Tables 3 and 5-7 are reported measured isotope effects for four systems-the water reaction of the a-fluoride,the bimolecular reaction of the a-fluoride with azide ion, the water reaction of the @-fluoride,and then the succinate-buffer-catalyzed reaction of the p-fluoride. Two aspects of the validity of these data, measured by the isotopic quasi-racemate method, need to be addressed, however. The first concerns mutarotation of the sugar. The method in its simple form assumes that the equilibrium composition of the anomers is reached with a time constant fast compared to hydrolysis. The data in Table 2 indicate that this is so for all reactions except that of the water reaction of the @-fluoride,where hydrolysis is only 10-fold slower than mutaro-
1560 J. Am. Chem. SOC.,Vol. 116, No. 17, 1994
Zhang et al.
Table 3. Kinetic Isotope Effects for the Neutral Hydrolysis of tation. On the worst-case assumption, that the initial hydrolysis a-Glucosyl Fluoride in 0.2 M Sodium Succinate Buffer, pH 6.0, I = yields a-glucopyranose exclusively, this will introduce an error 1.0 M (NaClOd) at 80.0 OC of about 10% of the effect on observed effects. site and isotope KIE (ua,U K l d average OMe EIEb The very large difference in rate between a- and 8-glucosyl fluorides (see Table 1, ref 24) made it impracticable to measure CY-D 1.135 (0.0004,0.020) 1.142 1.137 1.188 kinetic isotope effects at the sametemperature for both a n o m e r ~ . ~ ~ 1.138 (0.0003,0.026) *0.0075 1.144 (0.0002,0.017) We therefore measured isotope effects on the hydrolysis of the 1.152 (0.0002,0.020) ,&anomer at 50 O C and of the a-anomer at 80 O C . The effects 6-D 1.072 (0.0002,0.010) 1.067 1.073 for the a-anomer can thus be compared directly with those 1.065 (0.0003,0.014) a0.0077 previously reported for methyl a-and 8-glucosides? If the effects 1.069 (0.0002,0.009) areclassicalzero-point energy effects, then Tln(kL/ks)= constant 1.055 (0.0002,0.014) and the effects for the &anomer should be reduced by 10% of y(CS)-D 0.983 (0.0002,0.0017) 0.979 0.977 0.976 (0.0002,0.0006) h0.0032 the effect for proper comparison; if there is a contribution to the 0.978 (0.0003,0.0070) heavy atom effects from the isotope effect on molecular moments 0.971 (0.0002,0.0050) of inertia, then the temperature dependence will be less steep. 1-13c 1.030 (0.0007,0.0039) 1.032 1.007 1.004 Leaving group '80 effects on the acid-catalyzed hydrolyses of 1.033 (0.0009,0.0042) A0.0032 oxygen glycosides,8J+26 as well as @Ir values of approximately 0 1.031 (0.0007,0.0020) for the acid-catalyzed hydrolyses of glycosides and of below -1 1.037 (0.0013,0.0062) 5-180 0.981 (0.0005,0.0020) 0.984 0.9965 0.978 for the spontaneous hydrolyses of 2-(aryloxy)tetrahydropyran~~~ 0.982 (0.0010,0.0054) &0.0049 and 2-(aryloxy)tetrahydrofurans28 establish that the transition 0.990 (0.0006,0.0026) state for hydrolysis of oxygen glycosides is very late, whether the leaving group is protonated in a preequilibrium step or not. j31e a For the acid-catalyzed hydrolysis of methyl a-glucoside, taken from ref 8. * Calculated for the appropriate rotamer of dihydroxymethane by values for the spontaneous hydrolyses of glycosyl pyridinium I. H. Williams and quoted in ref 8. ionsl2.29 of below -1 and large leaving group 15N kinetic isotope effects for nucleoside hydrolysis30 likewise establish that at the Table 4. Kinetic Isotope Effects on the Reaction of a-Glucosyl transition state for hydrolyses of these N-glycosides the C-N Fluoride with Azide Ion (0.2 M Sodium Succinate Buffer, 2.0 M bond is largely broken. Since there is such ample experimental Sodium Azide, pH 6.0 at 50.0 "C) evidence for the lateness of the transition states for glycosylKIE (um ~ K I E ) oxygen and glycosyl-nitrogen bond cleavage, we assumed similar, site and isotope 0.2 M SU/2 M NaNn, 50 "C average almost complete, carbon-fluorine cleavage at the transition state a-D 1.178 (0.0006,0.014) 1.169 for glycosyl fluoride hydr01ysis.I~ 1.162 (O.OOO5, 0.017) +0.0082 Water and Azide Reactions of a-Clucosyl Fluoride. Isotope 1.167 (0.0005,0.014) 1-13c 1.087 (0.0003,0.026) 1.085 effects in the water reaction of a-glucosyl fluoride are set out in 1.092 (0.0004,0.017) a0.0082 Table 3, where they are compared with those for the acid-catalyzed 1.076 (0.0008,0.014) reaction of methyl a-glucoside. Whereas the secondary deuterium effects for the two systems are broadly similar, the I3C and I8O effects are radically different. The I3C effect is indeed totally phosphate: this corresponds to a I3C effect of around 0.5%. incompatible with a unimolecular reaction. S N reactions ~ are Schramm's group has reported a somewhat higher effect for the commonly associated with low values of reaction center carbon cleavage of a glycosyl-nitrogen bond, k12/k14of 1.049 f 0.010 isotope effects, because the loss of zero-point energy occasioned for acid-catalyzed hydrolysis of AMP.33 The much higher 13C by the cleavage of the carbon leaving group bond is largely offset effect for the water reaction of a-glucosyl fluoride than for the by the increase in zero-point energy associated with double bond acid-catalyzed reaction of methyl a-glucoside thus provides telling character in the bonds connecting the reaction center to groups support for the conclusion of Banait and Jencks? that the which stabilize positive charge, whereas S N reactions ~ are glucopyranosyl cation is too unstable to exist in contact with an associated with high effects. Thus, for example, as the nucleoanionic leaving group (such as fluoride) but in the presence of philicity of the solvent is decreased in the solvolysis of isopropyl a neutral one (such as methanol) it has a real existence. 2-naphthalenesulfonate from ethanol through to hexafluoro-2propanol, the reaction center I4Ceffect falls from 1.095to 1.044." The question of bimolecularity of the reactions of a-glucosyl In respect of glycoside hydrolysis, in addition to our effects with fluoridewas probed further by running the reaction in 2 M sodium the acid-catalyzed hydrolyses of methyl glucosides,8 Goiten et azide (Table 4), conditions which are known to result in direct 01.32 report a value of kl2/kl4 of 1.010 f 0.006 for the a-carbon displacements to give /3-glucosyl azide? the lower temperature effect on the loss of pyrophosphate from 5-phosphoribosyl pyro(50 "C) was an inescapable instrumentalconsequence of the faster reaction under these circumstances. The reaction center carbon (23,) For maximum accuracy of the individual readings, the polarimeter effect is very high indeed (1.OS5 f 0.008): we present an example requires a 10-min integration time and the minimal requirement for a data of the primary data in Figure 3. This effect is not however set is around 40 points extending over four half-lives. A minimum run time unprecedented: Ando et 0 1 . report ~~ I4C effects of 1.1 2-1.16 on of 5-6 h is thus required for the faster reacting isomer. (24) (a) Banks, B. E.C.; Meinwald,Y.; Rhind-Tutt, A. J.; Sheff,I.;Vemon, the reaction of N,N-dimethylanilines with benzyl arenesulfonates, 1961,3240. (b) Rosenberg, S.;Kirsch, J. F. Biochemistry C. A. J . Chem. SOC. and Wong and Sch0wen,3~a I3C effect of 1.08 on the reaction 1981, 20, 3196. of methoxide with S-methyldibenzothiophenium ion. The azide (25) Bennet, A. J.; Sinnott, M. L.; Wijesundera, W. S.S.J. Chem. Soc., Perkin Trans. 2 1985, 1233. reaction is associated with a marginally higher a-deuterium effect (26) Bennet, A. J.; Davis, A. J.; Hosie, L.; Sinnott, M. L. J. Chem. SOC., than the water reaction, which could be a temperature effect. Perkin Trans. 2 1987, 581. Azide, though, is a "softer" nucleophile than water, and it was (27) Craze, G.-A,; Kirby, A. J. J . Chem. Soc., Perkin Trans. 2 1978,354. (28) LBnnberg, H.; Pohjola, V. Acra Chem. Scand. 1976,30A, 669. found in the nucleophilic reactions of N,N-dimethyl-N-(meth(29) Jones, C. C.; Sinnott, M.L.; Souchard, I. J. L. J . Chem. Soc., Perkin Trans. 2 1977, 1191. (30) Parkin, D. W.; Schramm, V. L. Biochemistry 1987, 26, 913. (31) Yamataka, H.; Tamura, S.;Hanafusa, T.; Ando, T. J. Am. Chem. SOC. 1985, 107, 5429. (32) Goiten, R. K.; Chelsky, D.; Parsons, S.M. J. Biol. Chem. 1978,253, 2963.
(33) Parkin, D. W.; Leung, H. B.; Schramm, V. L. J . Biol. Chem. 1984, 259,9411. (34) Ando, T.; Tanabe, H.; Yamataka, H. J. Am. Chem. SOC.1984,106, 2084. (35) Wong, 0.S.-L.; Schowen, R. L. J . Am. Chem. Soc. 1983,105,1951.
Hydrolyses of a- and 8-Glucopyranosyl Fluorides
-
0.01 0
v
8
0.008
0.006
J. Am. Chem. SOC.,Vol. 116, No. 17, 1994 7561
Thedihedral angle H242-CI-Fcan likewise beapproximately located from the 8-deuterium kinetic isotope effect. &Deuterium kinetic isotope effects have their origin in two phenomena, hyperconjugation and the inductive effect of deuterium, and can be considered in terms of eq 6.37 ln(kHkD) = cosz e ln(kH/kD)max + ln(kH/kD),
(6)
The first term of eq 6 represents hyperconjugation of the C-L cr orbital with an electron-deficientp orbital on an adjacent carbon
atom: 0 is the dihedral angle between the C-L bond and this orbital. ln(kH/kD),, is the maximal hyperconjugative effect obtained when the C-L bond and the p orbital are exactly eclipsed and increases as the positive charge on the adjacent carbon atom 0 1.0 lo4 2.0 lo4 3.0 IO4 4.0 lo4 increases, with the associated weakening of the C-L bond. The Time ( s ) second term in eq 6 represents the small, electron-releasing Figure 3. Time course of optical rotation of a solution containing a-Linductive effect of deuterium. glucosyl fluoride and [l-13C]-a-~glucosylfluoride in 0.2 M sodium The data of Kresgeand Weeks38onsecondary deuteriumisotope succinate, pH 6.0,2.0 M sodium azide at 50 OC (u. = 0.0004, UKIE = effects on the hydrolyses of acetaldehyde diethyl acetal and of 0.0 173). ethyl vinyl ether at 25 OC provide a basis for estimating the values of both h(kH/kD)m,x-and ln(kH/kD)maxand ln(kslkD)i oxymethy1)aniliniumions7bthat "softer" nucleophiles gave bigger for reactions which generate an oxocarbonium ion in water. In a-deuterium kinetic isotope effects. water, the hydrolysis of the acetal is specific acid catalyzed, The very different effects arising from 1 8 0 substitution in the whereas the rate-determining step in the hydrolysis of ethyl vinyl case of the fluoride (0.984 f 0.005) compared to the methyl ether is protonation of the methylene carbon. Both hydrolyses glycoside (0.9965) suggest immediately that doublebond character involve the ethoxyethyl cation as a real intermediate. The between 05 and C1 is much more pronounced at the transition inductive effect of a-deuterium substitution can be estimated state for fluoride hydrolysis than for methyl glucosidehydrolysis. from the isotope effect on the hydrolysis of CH,=CL-OEt, The low value of the ring 180effect in the case of the methyl since there is no change of hybridization at the carbocationic a-glucoside was the main determinant of our finding that the carbon and the effect of (0.965) is wholly inductive. Inductive transition state for this compound was derived from the IS3 skew effects are attenuated by a factor of approximately 2.5 per conformation: qualitatively, therefore, a different reactive additional carbon atom between the reaction center and the site conformation is anticipated for the fluoride. of s~bstitution,3~ so we take h(kH/kD)i as -0.014. The hyperConformational information can be obtained from the inverse conjugative8-deuterium kinetic isotope effect of a freely rotating ring 1 8 0 isotope effect by the following argument. Effective cL3 group is 3/2(k~/k~)max,40 and the experimental @-deuterium delocalization of charge and, hence, development of 0 5 4 1 double kinetic isotope effect for the hydrolysis of CL&H(OEt), is 1.134 bond character is possible only with the use of the p-type lone f 0.011.38 By using a value of -0.014 for ln(kH/kD)f, we can pair on oxygen: the sp-typelone pair is largely ineffecti~e.~~ Since estimate h(kH/kD),x as 0.112 at 25 OC. Correction of this the strength of p p orbital overlap depends on the square of the estimate of In(kH/kD),x to 80 'C assuming nn(kL/kH) = cosine of the dihedral angle between the two p orbitals, we can constant and taking h(kH/kD)i, as -0.014 allows 8 for the write eq 5, (kl6/kl8)max is the isotope effect when the p-type lone hydrolysis of a-glucosyl fluoride to be estimated from eq 6. The resulting value of 0 of 25' (Le., a dihedral angle H2-C2-Cl-F of 155') is obtained. These estimates of 0 and w and ordinary qualitative conformational reasoning indicate that the transition state for a-glucosyl fluoride is 4C1chair flattened toward 4H3. pair and the electron-deficientp orbital on C 1 are exactly aligned, The two other secondary deuterium effects are also consistent and o is the dihedral angle between the p-type lone pair on oxygen with this picture. The simplest possible interpretation of the and the electron-deficient p orbital on C1. In the work on the a-deuterium effect, that the degree of rehybridization at C1 is acid-catalyzed hydrolyses of the methyl glycosides? two extreme merely ln(aKIE)/ln(EIE), leads to theconclusionthat theanomeric estimates of (k16/k18)mpx were used. We argued that any kinetic carbon is 74% rehybrized. Whatever the corrections required effect would be less than the equilibrium effect for oxocarbonium for the bimolecularity of the reaction, estimation of the bond ion generation and based one estimate on Williams' theoretical angles defined by the three nondeparting substituents and the calculations for the dehydration of dihydroxymethane, shown in anomeric center as 117' at the transition state will introduce Table 1. At the other extreme, we took the kinetic effect of 0.988 for the acid-catalyzed hydrolysis of isopropyl [ I-lEO]-a-~- errors of only a few degrees. Likewise, the magnitude of the r(C5) effect qualitatively supports the contention that there is arabinofuranoside,zs which takes place by a ring opening no large conformational change between the ground state and mechanism, in the rate-determining transition state of which the transition state. The C5 effect arises in principle from two ring oxygen is protonated and the C1-01 bond has double bond phenomena, the inductive effect of deuterium and the deuterium character. However, the measured inverse effects for the conformational isotope effect. Whereas the latter is modest in hydrolyses of glucosyl fluorides are bigger than those for this the case of cyclohexane derivatives (the deuterium prefers the system, so that here 0.988 is a poor measure of (kl6/k18)max. equatorial orientation by about 6 cal the presence of an Despite the complication caused by the bimolecularity of the a-heteroatom can cause it to be quite ~ubstantial.~, Existing reaction, therefore, the theoretical value for the equilibriumisotope effect calculated for the conversion of the appropriate rotamer (37) Melander, L.; Saunders,W. H. Reaction Ratesoflsotopic Molecules; John Wiley and Sons: New York, 1980. of dihydroxymethane to the protonated formaldehyde cation is (38) Kresge, A. J.; Weeks, D. P. J . Am. Chem. SOC.1984, 106, 7140. an attractive zeroth approximation to (kl6/kl8)max. On this basis (39) See, e.g.: Isaacs, N. S. Physical Organic Chemistry; Longmans: o is estimated to be 32O. Harlow, U.K., 1987; pp 144. (36) Briggs, A. J.; Evans, C. M.; Glenn, R.; Kirby, A. J. J. Chem. SOC., Perkin Trans. 2 1983, 1637.
(40) Hehre, W. J. Acc. Chem. Res. 1975, 8, 369. (41) (a) Anet, F. A. L.; Kopelevich, M. J. Am. Chem. Soc. 1986, 108, 1355. (b) Anet, F. A. L.; O'Leary, D. J. Terrahedron Lett. 1989,30, 1059.
Zhang et al.
7562 J. Am. Chem. Soc., Vol. 116, No. 17, 1994
n
Table 5. Comparison of Experimental Kinetic Isotope Effects with Those Calculated from the Transition State of Figure 4 by BEBOVIB-IV site and isotope of substitution a-D
8-D 5-D
I-*3c
5-'*0
KIE(expt1)b
KIE(theor)C
1.142 1.067 0.979 1.032 0.986
1.142 1.070 0.979 1.023 0.989
\.\.\. 15%; ..,..,, .....$,$... ?,.A
.\.,.I.\.
~
bond order (bond) 0.961 (CI-HI) 0.844 (C2-H2) 0.9 18 (C5-H5)
1.920 (C1-05) 0.001 (Cl-F1) 0.001 (CI-OH2)
Table 6. Kinetic Isotope Effects for the Hydrolysis of 8-Glucosyl Fluoride in 0.3 M Bis-Tris Buffer, pH 6.0, at 50.0 "C, I = 1.0 M (N aC104)
site and isotope
a-D
d
8-D 1-13C
5-'*0
KIE (um bKIE)
average
-OM@
CalCd EIEb
1.085 (0.0003,0.0047) 1.085 (0.0004,0.0055) 1.087 (0.0006,O.O 14) 1.021 (0.0005,0.0027) 1.033 (0.00 10,0.0078) 1.037 (0.0008,0.0037) 1.020 (0.0004,O.OO 14) 1.O17 (0.0005,0.0026) 1.O16 (0.0004,0.00 1 2) 0.982 (0.0006,0.0019) 0.99 1 (0.0006,O.OO1 5) 0.983 (0.0007,0.0018)
1.086 10.00 12
1.089
1.137
1.030 f0.0083
1.045
1.017 10.0022
1.011
1.00
0.985 f0.0049
0.991
0.976
Figurer(. BEBOVIB-IV transition-state structure for a-glucosyl fluoride hydrolysis.
For the acid-catalyzed hydrolysis of methyl a-glucoside, taken from ref 8. Calculated for the appropriate rotamer of dihydroxymethane by I. H.Williams and quoted in ref 8.
data (on 1,3-dio~olans)~~ suggest that the equatorial orientation of deuterium a- to a single ring oxygen should be favored by around 24 cal mol-'. This by itself could account for an inverse y-deuterium effect of 34-76. If there were a large change in conformation, this effect would be superimposed on an inductive effect in the same sense, arising from charge buildup on the ring oxygen,which is known to be significant from the ring l 8 0 effects. Qualitatively, therefore, the effects are all consistent with a bimolecular attack of water on a-glucosyl fluoride, through an exploded transition state immediately derived from the groundstate 4C1conformation of the substrate. This is in stark contrast to the transition state with the neutral leaving group methanol,8 which does not involve a nucleophilic water and in which the conformation of the sugar ring is derived from the conformation. Estimation of the Transition-State Structure for the Water Reaction of a-Glucosyl Fluoride Using BEBOVIB-IV. Tanaka et al. (Tanaka, Y.; Tao, W.; Blanchard, J. S.; Hehre, E. J. J. Biol. Chem., under review) have recently described isotope effects for enzymic hydrolyses of a-glucosyl fluoride and have derived transition-state structures using BEBOVIB-IV. By starting with a theoretically-refined structure, bond orders were varied systematically and the resulting calculated isotope effects compared with experiment. Using their input parameters and procedure, we were able to locate the transition state for a-glucosyl fluoride hydrolysis shown in Figure 4, which reproduced the experimental effects reasonably well (Table 5 ) . Particularly noteworthy is the H2-C2-CI-F dihedral angle, which in Figure 4 is 156': the approximate calculation based on the @-deuteriumeffect using reactions which generated the ethoxyethylcation as a model (vide supra) gave 155'. However, while the inverse ring I80effect was successfully modeled in this system, we found that systematic variation of the OS-C1 bond order gave inverse ring I8Oeffects
only over a very narrow range (1.8 < bond order < 2.0). The value of w derived from the transition state of Figure 4 is 12O, as compared with 32' estimated empirically. The (comparatively minor) disagreement may be a consequence of the -30% experimental errors on the ring 1 8 0 effect and its cosine squared dependence on w. A transition state for the water reaction of @-glucosylfluoridecouldnot be located using the parameterization of Tanaka et al. for the a-fluoride. Water Reaction of &Clucosyl Fluoride. In Table 6 are set out isotope effects for the water reaction of this compound at 50.0 OC. Comparison with theeffects for theacid-catalyzed hydrolysis of the methyl glucoside reveals no major differences. Particularly noteworthy is the small reaction center carbon effect: change of the leaving group from neutral MeOH to anionic F- causes no great change in this effect, unlike the situation in the a-series. This lower reaction center carbon effect for the &fluoride as compared to the a suggests a lower degree of nucleophilic assistance by water to the departure of fluoride. It is tempting tospeculate that this may be related to the long-recognized greater susceptibility of axial leaving groups to bimolecular nucleophilic displacement than equatorial leaving groups.43 The glucopyranosy1 cation may be so finely on the border of a real existence that it is too unstable to exist with an anionic nucleophile on the a-face, but not on the @-face. Estimation of the dihedral angles w and 6,from the ring 180 effect and @-deuteriumeffect in the same way as for the a-fluoride, with a temperature correction based on the assumption that 7ln(kl/kh) = constant, gives w = 41' and 0 = SO'. Qualitatively, these angles are consistent with reaction through a flattened 4CI chair, as observed for methyl @-glucoside.8 Although the possibility of reaction through nonchair conformers cannot readily be dismissed on the basis of these angles alone, if the @-fluoride
(42) (a) Anet, F. A. L.; Kopelevich, M. J . Am. Chem. Soc. 1986, 108, 2109. (b) Anet, F. A. L.; Kopelevich, M. J . Chem. Soc., Chem. Commun. 1987,595. (c) Carr, C. A.; Ellison, S. L. R.; Robinson, M. J. T. Tetrahedron L m . 1989, 30, 4585.
(43) (a) Bone, J. A,; Pritt, J. R.; Whiting, M. C.; J . Chem. Soc., Perkin Trans. 2 1975, 1447. (b) Storesund, H. J.; Whiting, M. C. J . Chem. Soc., Perkin Trans. 2 1975, 1452. (c) Schadt, F. L.; Bentley, T.W.; Schleyer, P. v. R. J . Am. Chem. Soc. 1976.98.7667. (d) Bentley, T. W.; Bowen, C. T. J . Chem. Soc., Perkin Trans. 2, 1978, 557.
Hydrolyses of a- and 8- Glucopyranosyl Fluorides Table 7. Kinetic Isotope Effects on the Hydrolysis of @-Glucosyl Fluoride in 0.3 M Sodium Succinate Buffer, pH 6.0, I = 1.0 M (NaCIO,), at 50.0 OC
site and isotope
KIE (ne,~ K I E )
average
a-D
1.101 (0.0005,0.0067) 1.100 (0.0004,0.024) 1.109 (0.0006,0.017) 1.058 (0.0003,0.007 1) 1.059 (0.0004,0.0090) 1.060 (0.0003,0.0070) 0.973 (0.0005,0.0014) 0.977 (0.0005,0.0019) 0.987 (0.0005,0.0014) 0.988 (0.0006, 0.0013) 1.073 (0.0009,0.0070) 1.056 (0.0009,0.0049) 1.063 (0.0012,0.0049) 1.064 (0.0009, 0.0089) 0.985 (0.0003, 0.0013) 0.992 (0.0003,0.0014) 0.987 (0.0003, 0.0013)
1.105 k0.0053
8-D 5-D
1-13c
5-'*0
J. Am. Chem. SOC.,Vol. 116, No. 17, 1994 7563
--k
0.0015
2
0.0012
c
AN
0
1.059 fO.OO1O 0.981 10.0074
0.0°06
t
t
0~0003
1.064 10.0070 0.988 10.0036
reacted through a boat conformation and the a-fluoride through a chair, then access of an nucleophilic water molecule to the reaction center for both substrates would be comparably ready and the interpretation of the reaction center 13C effects would become problematic. The a-deuterium effect is lower than for the a-fluoride but not as a fraction of the calculated equilibrium isotope effect: the approximation that the isotope effect is a linear function of the degree of rehybridization indicates that the reaction center is 66% rehybridized. Succinate Reaction of fi-Clucosyl Fluoride. In Table 7 are given isotope effects for the succinate reaction of @-glucosyl fluoride. The obvious interpretation of the results, supported at first sight by the very high reaction center l3C effect, that one is observing the action of succinate as a nucleophile, on closer inspection fails to rationalize the facts. Most noticeably, it appears as succinate monoanion rather than succinate dianion, which is the more effective catalyst of the reaction (Figure 5 ) . Additionally, ascription of buffer catalysis to a nucleophilic reaction encounters the problem why the catalysis should be much more marked for the 8-fluoride than for the a , when the backside approach of a nucleophile is so much easier in the a-case. A further, circumstantialpieceof evidence is that the rotation change on reaction with succinateis thesame as that for the water reaction, so the product is glucose, not the acylal. Bifunctional catalysis by succinate monoanion, in which -COOH acts as a general acid and -COO- as a nucleophile is sterically disfavored (the reaction would correspond to a 9-endotet cyclization). A transition state which explains the facts is one in which most of the reaction occurs by neutral succinic acid acting upon the C2 oxyanion of the substrate (Figure 6). Such a transition state explains the high reaction center l3C effects (since the key step is an intramolecular nucleophilic displacement by the C2
0' 0
O I
0.2
0.4
0.6
0.8 1 Acid Form(%)
Figure 5. Dependence of the buffer catalyticconstant for the hydrolysis of @-glucosylfluoride in 0.3 M succinateat 50 O C , I = 1.OM by NaC104
upon composition of the buffer. The buffering ionization is the second ionization of succinic acid, so that the acid form is succinate monoanion and the basic form succinate dianion.
Figure6. Qualitativetransition-statestructure for thesuccinate-catalyzed hydrolysis of &glucosyl fluoride. oxyanion) and also the greater susceptibility of the equatorial j3-fluoride to this form of catalysis. It permits a small amount of catalysis, formally by succinate dianion, occurring through a similartransition state involving generalacid catalysis by succinate monoanion of the ring closure of the substrate conjugate base. The existence of a preequilibrium, moreover, rationalizes the higher &deuterium effect than for the anion. Formation of the substrate anion will be disfavored inductively by the 8-deuterium: this phenomenon could indeed account for the whole of the observed effect.44 Enhanced reactivity upon ionization of a 2-hydroxy group in the ribofurano series has been shown to be associated with the persistence of high a-deuterium kinetic isotope effecw45 these effects in the present system are therefore consistent with the transition state of Figure 6.
Acknowledgment. We thank the NIH for Grant GM 42469, Dr. John Blanchard and Dr.Wen Tao for assistance with the BEBOVIB calculations, and Professor W. P. Jencks for preprints of ref 9 and helpful comments on the manuscript. (44) Williams, I. H. J. Chem. Soc., Chem. Commun. 1984, 1069.
(45) Johnson, R. W.; Marschner, T. M.; Oppenheimer, N. J. J. Am. Chem. SOC.1988, 110, 2257.
